Milan Segedinac, Goran Savić and Zora Konjović
Faculty of Technical Sciences, University of Novi Sad, Fruskogorska 11, Novi Sad, Serbia
Keywords: Knowledge representation, Curriculum development, e-learning, Learning objectives, Instructional design.
Abstract: This paper presents the formal representation of knowledge used in curriculum development process. Four
curriculum components are represented separately: learning objectives, learning experiences, the
organization of learning experiences and evaluation of learning outcomes. Learning objectives are formally
represented using ontologies. Learning experiences consist of learning objects and achievements assessment
instruments (tests) and they are specified using IMS Content Packaging standard. Learning experiences are
mapped to the learning objectives ontology using XML. For describing instructional design, we proposed a
special-purpose language implemented using XML notation. The achievement of a learning objective is
assessed using test items linked to this particular objective. Such an approach allows more flexible
management of the curriculum as a whole and easier modification of the particular components than in
classical approach.
For the proper management of a curriculum
development process, a formal specification of all
curriculum components is essential. This formal
representation should be used for automation of the
curriculum creation process. Besides, it enables
automation of curriculum. We start by analyzing
Tyler’s rationale for curriculum development.
Tyler’s rationale states that in any educational
setting, when developing curriculum, four
fundamental questions must be considered (Tyler,
1. What educational purposes should the
school seek to attain?
2. What educational experiences can be
provided that are likely to attain these
3. How can educational experiences be
effectively organized?
4. How can we determine whether these
purposes are being attained?
These questions can be reformulated in a more
familiar way as forming the process of curriculum
and instruction development consisting of: the
selection of learning objectives, the selection of
learning experiences, the organization of learning
experiences, and the evaluation of learning outcomes
(Tyrell, 1974).
In this paper it is shown how the knowledge used
in the steps of the process of curriculum and
instruction development can be represented. The
relationships among the steps are stressed as well.
Each step is represented as an independent
component, because such approach is more flexible
and it provides easier modification of the
The formulation, classification and organization of
learning objectives play very important roles in
Tyler’s rationale since all other steps proceed from
them. In this paper, a learning objective is
understood as “an explicit formulation of the way in
which students are expected to be changed by the
educational process”, and these changes can be
represented using taxonomies (Bloom, Engelhart,
Furst, Hill and Krathwohl 1956).
When the content of learning objectives is
represented, the number of the taxonomies to be
represented equals to the number of courses. The
explicit representation of the content facilitates
Segedinac M., Savi
c G. and Konjovi
c Z..
DOI: 10.5220/0003052303270330
In Proceedings of the International Conference on Knowledge Engineering and Ontology Development (KEOD-2010), pages 327-330
ISBN: 978-989-8425-29-4
2010 SCITEPRESS (Science and Technology Publications, Lda.)
establishing the relationships among the learning
When representing learning objectives, students’
behaviours involved in these objectives should also
somehow be represented. For that purpose Bloom’s
taxonomy, taxonomy of students’ behaviours which
represent the expected learning outcomes had been
designed (Bloom et al., 1956).
Three levels of specificity of learning objectives
are identified in (Anderson, Krathwohl, Airasian,
Cruikshank, Mayer, Pintrich, Raths and Wittrock,
2001.): global, educational, and instructional.
Although ontologies have wider range of use,
they are natural tool for representing taxonomies.
Therefore we have decided to use OWL full to
represent learning objectives. Learning objectives
ontology conjoins domain knowledge taxonomy,
Bloom’s taxonomy, and learning objective
specificity taxonomy. In Figure 1 an example of
representation of a concrete learning objective is
h a s i n sta nc e
Figure 1: Representation of a learning objective
While learning, a student must achieve less
complex learning objectives before he or she can
proceed to more complex ones. Accordingly, the
feasible (consistent) combinations of the learning
objectives have to be established. Knowledge space
theory facilitates representing feasible individual
knowledge structures with respect to the set of
problems that a student should be able to solve
(Doignon and Falmagne, 1999). As achieving a
learning objective means being able to solve a
problem, the set of problems can be mapped to the
set of learning objectives (Segedinac, Savic and
Slivka. 2010).
When representing learning objectives, we
confine to learning spaces with the quasi ordinal set
of learning states. These knowledge spaces can be
represented using surmise relation introduced in the
set of learning objectives. Learning objective q
surmises learning objective q
if, from knowing that
a student has achieved q
we can infer that he or she
has achieved q
(Doignon and Falmagne, 1999).
Organizing learning objectives in a knowledge space
facilitates knowledge assessment and allows
chaining of learning objectives.
According to Bloom’s taxonomy (Bloom et al,
1956), when learning objectives refer to the same
domain knowledge, the objectives which require
lower cognitive processes are prerequisites for those
which require higher cognitive processes. This rule
was built in our model using SWRL rules. Besides
that, user can specify surmise relation among
learning objectives freely.
test items and learning objects
mapping on learning
objectives ontology
Figure 2: The mapping of learning objects and test items
to learning objectives chain with surmise relation.
After knowledge space was built on the set of
learning objectives, a chain of learning objectives
should be defined. For that purpose we introduced
relation next (a relation of total ordering) in the set
of learning objectives, forming a chain of learning
objectives. This relation allows us to suggest the
order of achieving learning objectives and is useful
when organizing learning experiences. Relation next
was introduced with respect to surmise relation,
specifying new SWRL rules. In Figure 2, an
example of learning objective chain with surmise
relation and learning objects and corresponding test
items is presented.
A student gets most of the learning experiences by
consuming specific learning resources. Thus, our
component for representing learning experience
should formally describe learning resources. For this
purpose, we suggest IMS Content Packaging
specification, which is globally accepted.
Although we define the learning objectives and
learning material as separate components, there is
KEOD 2010 - International Conference on Knowledge Engineering and Ontology Development
still strong relationship between these components.
Students use learning objects to achieve learning
objectives and teachers use tests to evaluate
students’ knowledge. Thus, learning resource is
always related to one or more learning objectives.
To describe this relationship, we have created an
intermediate component that defines mapping
between learning resources and learning objectives.
We use a particular XML document to define this
mapping. The relationship between learning
objectives and material is shown in Figure 2.
It can be noticed that in our approach ontology
of learning objectives has a role to define relations
among learning resources. Thus, the relation is not
defined as a part of the resource, but it is implicitly
defined through ontology. By this, it is easier to
change relation between two resources, i.e. in order
to change relation between two resources; one
should only map the resource to another objective.
Likewise, when adding a new resource, its
relationships with other resources are indirectly
defined by its learning objectives.
In the previous chapter we assumed that learning
experience is closely related to a specific learning
resource. Thus, the organization of learning
experiences is actually defined by two components -
selecting and organizing learning resources. These
two components define instructional design used in a
course. Although there are numerous different
instructional strategies (Ryder, 2010), there is no
formal language aimed at specifying instructional
design. Formal specification of instructional design
should enable computer-aided reasoning about
instructional design. For that purpose, we have
created an XML-based language formally describing
instructional design used in the course. This
language should provide ability for a teacher to
define the order of learning activities and criterion
for selecting learning resources (number of
resources, their type, and priority). For these
reasons, our instructional design language defines
the path through the learning objectives (objectives
are represented as nodes in the ontology, see Figure
2). For each objective, the language specifies an
ordered subset of the resources mapped to the
In Figure 3. the UML model of our proposal for
instructional design specification is presented.
Root element in the model is instructional design
element representing the course. Structure element is
a generic learning element in organization. There are
two different types of elements – sequence and
learning object. Sequence is a chain of other
elements. Learning object is a unit of learning on the
lowest hierarchical level. It actually represents a
concrete learning resource. For each sequence we
can define a specific strategy for selecting learning
resources. This strategy is defined in selection rule
element. Selection rule aggregates two lists of
Object selection elements. First list contains objects
that are included in the course. The second one is for
excluded objects. Object selection element specifies
learning objects for including or excluding. For
included learning objects, we set an integer value
called priority, which defines the order of learning
objects in the sequence.
Besides the course structure, sometimes it is
necessary to define relationships among learning
elements. For example, in mastery learning, a
student can’t proceed to the next learning objective
until he or she has completed the previous one. So,
we need to define the relationship between two
learning objectives. These relationships are specified
using element relation element. This element has
references to the source and destination learning
elements, respectively.
Condition specifies when two elements are in the
relationship. If the condition is satisfied, a specific
action (defined in the then action element) is done.
Otherwise, an action defined in the else action
element is executed.
1 0..*
instructional design
structure element
learning object
element relation
selection rule
object selection
then action
else action
Figure 3: The UML 4model of proposed instructional
design specification.
On the basis of the described UML model, we
created an XML schema. Listing 1 presents a part of
an XML document, created according to the schema.
The document formally describes instructional
design used in the “Web programming” course held
at Faculty of Technical Sciences Novi Sad in 2009.
The precise identification of the set of achieved
learning objectives plays the key role in successive
learning, because it leads further learning process
(Mager, 1984).
Knowledge space formed on the set of learning
objectives proposed in this paper allows us to use
techniques enabling the explicit specification of
achieved learning objectives. Each test item is
mapped to specific learning objective(s) and
multiple test items can be mapped to the same
learning objective (Figure 2). These techniques
explicitly identify the set of learning objectives that
student has achieved, and can be used in interactive
assessment (Degreef, Doignon, Ducamp and
Falmagne, 1986; Falmagne and Doignon, 1988), as
well as in classical educational settings (Segedinac
et. al. 2010).
Listing 1: Instructional design example.
In this paper a formal knowledge representation
model for curriculum development process
automation is proposed. The model consists of four
components: learning objectives, learning
experiences, the organization of learning
experiences, and the evaluation of learning
outcomes. Classical approach to modelling
curriculum development process often uses the
monolithic representation of the process resulting in
situation where small changes cause alteration of the
whole structure. In our approach, each component is
modelled separately which allows managing
curriculum in a more flexible manner and altering
components more easily than in a classical approach.
Future works will include extending one of the
existing open-source e-learning systems with
proposed curriculum development module. Such an
e-learning system would allow further pedagogical
research related to optimization and evaluation of
the educational process.
This paper was supported by the Ministry of science
and technological development, Republic of Serbia.
Anderson, L., Krathwohl, D., Airasian, P., Cruikshank K.,
Mayer R., Pintrich P., Raths J., Wittrock, M., 2001. A
Taxonomy for Learning, Teaching and Assessing A
Revision of Bloom’s Taxonomy of Educational
Objectives, complete edition. In TEMPLATE’06, 1st
International Conference on Template Production.
Bloom, B., Engelhart, M., Furst, E., Hill, W., Krathwohl,
D., 1956. Taxonomy of Educational Objectives The
Classification of Educational Goals, Handbook 1,
Cognitive Domain, Longmans, USA.
Degreef, E., Doignon, J., Ducamp, A., Falmagne, J., 1986.
Languages for the assessment of knowledge. In
Journal of Mathematical Psychology, 30. Academic
Press. Orlando.
Doignon J., Falmagne J., 1999. Knowledge Spaces,
Falmagne, J., Doignon, J., 1988. A class of stochastic
procedures for the assessment of knowledge. In British
Journal of Mathematical and Statistical Psychology,
41. British Psychological Society.
Mager, R., Lopes, J., 1984. Preparing instructional
objectives. David S. Lake. Belmont, 2
Ryder, M., 2010. Instructional Design Models, available
online at
itc_data/ idmodels.html
Segedinac, M., Savic, G., Slivka, J., 2010. Knowledge
State Identification: A Neural Network Approach. (in
Serbian) In Proceedings of Yu Info 2010, Kopaonik,
Tyler, R., 1949. Basic Principles of Curriculum and
Instruction, Chicago: University of Chicago Press.
Tyrrell, R., 1974. An Appraisal of the Tyler Rationale. In
The School Review, Vol. 83, No. 1, Responses to
"Symposium on Youth: Transition to Adulthood". The
University of Chicago Press.
<instructional-design root =”course”>
<sequence element = "go">
<sequence element = "io” >
<sequence element = "learning-object">
<include type="ec" priority="1"/>
<include type="exmp" priority="2"/>
<include type="all" priority="3"/>
<exclude type="exercise"/>
<learning-object type="exercise"/>
<sequence element="learning-object">
<include type="project" priority="1"/>
<learning-object label = "final_test"/>
KEOD 2010 - International Conference on Knowledge Engineering and Ontology Development